Ozone, hydroperoxides, oxides of nitrogen, and hydrocarbon budgets in the marine boundary layer over the South Atlantic

. The NASA GTE TRACE A mission sampled air over the South Atlantic and western Indian Oceans. Thirteen flight legs were flown within the marine boundary layer (MBL). The MBL was typically the cleanest air sampled (e.g., CH 4 < 1680 ppb, CO < 70 ppb, C2H 6 < 400 ppt, C3H 8 < 40 ppt, NOx < 15 ppt, and midday NO < 5 ppt) but was overlain by polluted air. The photochemistry of the MBL was influenced by oceanic emissions, surface deposition, and entrainment of pollutants from aloft. Chemical budgets were constructed for several species in the MBL in order to investigate these effects and are presented for ethane, ethylene, propane, propylene, n-butane, formic acid (HFo), methylhydroperoxide (CH3OOH), oxides of nitrogen (i.e., NO, NO2, PAN, HNO3), hydrogen peroxide (H202) , and ozone (03). A photochemical point model was used to evaluate local chemical production and loss. An entrainment model was used to assess material exchange between the lower free troposphere (FT) and the MBL and a resistance deposition model was used to evaluate material exchange across the air-sea interface. The results suggested the ocean to be the source of measured alkenes in the MBL and to be the most likely source of the shorter-lived alkanes: propane and n-butane. Ethane was the only hydrocarbon for which input from aloft may have exceeded its photochemical destruction. The estimated hydrocarbon sources from the ocean were in agreement with prior analyses. Transport from the lower FT together with surface loss could not account measured concentrations of HNO3, CH3OOH , H202, and CH20. CH20 significantly in the MBL.


The NASA GTE Transport and Atmospheric Chemistry near the Equator--Atlantic Experiment (TRACE A) was an investigation in to the causes of high tropospheric ozone (03)
concentrations observed over the South Atlantic in the months of September and October [Fishman et al., 1990 The remote MBL is recognized for its role in the destruction of tropospheric 03, the antithesis of the primary TRACE A mission. Lenschow et al. [1982], Kawa and Pearson [1989], and Paluch et al. [1994] evaluated the vertical flux of 0 3 at the top of the MBL and within the MBL. They showed the MBL over the Gulf of Mexico and the eastern North Pacific to be a sink for tropospheric 03. Ozone loss was postulated to be the result of photochemical destruction and ozone destruction/absorption at the ocean's surface.  examined 0 3 in and above the MBL over the North Atlantic in two Lagrangian experiments near the Azores. They could not resolve net 03 production from surface deposition and entrainment. Modeling studies by Liu et al. [1983] and Thompson et al. [1993] have shown the net destruction of ozone in the MBL to be photochemically mediated with 0 3 replacement from the overlying free troposphere (FT). These chemical models were applied to observations from the equatorial Pacific Ocean, where nitric oxide (NO) concentrations were less than a few ppt. Ayers et al. [1992] and Ayers et al. [1995] examined the correlation of 03 and H202 over diel and annual cycles at Cape Grimm, Tasmania, and showed the daytime destruction of 03 to be accompanied by a stoichiometric increase in peroxides (most likely hydrogen peroxide and a major fraction of methylhydroperoxide). They further suggested that replacement 03 was mixed into the MBL from aloft, while the peroxides were lost to the ocean's surface. The TRACE A measurements and photochemical model results permitted an evaluation of the rate and mechanism of net 03 production in the MBL over the South Atlantic under similar low NO conditions. Several investigations have attempted to evaluate the oceanic flux of hydrocarbons into the atmosphere and their impact on atmospheric chemistry, particularly 03 and hydroxyl (HO). A combination of atmospheric measurements, seawater measurements, air-sea exchange rates, and photochemical models have been used to infer hydrocarbon fluxes over the equatorial Pacific [Donahue and Prinn, 1993; Thompson et al., 1993], the North and South Pacific [Lamontagne et al., 1974], the Atlantic [Rudolph and Johnen, 1990;Plass et al., 1992], the Indian Ocean [Bonsang et al., 1988], and in the remote marine atmosphere [Donahue and Prinn, 1990]. Bonsang  Oxides of nitrogen along with hydrocarbons are central to defining oxidant species and their concentrations in the atmosphere. The remote MBL is one of the few regions of the world wherein noontime NO levels below 5 ppt are consistently found [McFarland et al., 1979;Torres and Thompson, 1993]. Zafiriou  While the deposition of HNO 3 to the ocean removes reactive nitrogen from the troposphere, thereby limiting net oxidant production, the downward flux of HNO 3 and the conversion of NO 2 to HNO 3 in the MBL has important biological implications. Biological productivity in the oligotrophic ocean is considered to be limited by the availability of fixed nitrogen and the atmosphere may be a source of this macronutrient [Duce et al., 1991]. The TRACE A oxides of nitrogen data allowed estimates of the flux of HNO 3 to the South Atlantic to be made and allowed a reexamination of earlier estimates of the air-sea input of fixed nitrogen over this region.
Hydrogen peroxide, CH3OOH, and CH20 measurements have been used to validate photochemical model odd hydrogen (e.g., HO, HO2, RO2) chemistry through the comparison of observed and predicted concentrations. The comparisons with airborne data above the boundary layer are in agreement within stated measurement and kinetic uncertainties [Jacob et al., this issue;Crawford et al., 1995]. However near the Earth's surface, models and measurements diverge with the models typically indicating higher values than observed [Lowe and Schmidt, 1983;Thompson et al., 1993;Crawford et al., 1995;Jacob et al., this issue], and surface deposition of the more highly soluble species, e.g., CH20 , HNO3, and H202, has often been invoked to explain these differences. The suite of the more soluble species measured in TRACE A (H202, CH3OOH, CH20 , HNO 3, and HFo) together with the measurements of 0 3 , NO, and hydrocarbons permitted us to examine this assumption and test for consistency between the deposition of species with short photochemical timescales (e.g., H202, CH3OOH, or CH20 ) and those with longer photochemical timescales (e.g., 03, HNO3 or HFo). These data and their analysis provide additional constraints on the models and, if successful, place greater confidence in other model products (e.g., HO or net 0 3 production).   The oceanic equivalent gas phase concentration in the upper ocean, C* /,AS, was assumed to be negligible for the species listed. Ki,AS was evaluated using the simplified thin-film parameterized model of Duce et al. [1991]. Air-sea transfer was assumed to be limited by atmospheric turbulence for all but CH3OOH, which has a significantly lower solubility in water , than the other species. Hence its uptake depends upon both an atmospheric transport velocity, k a, and an oceanic transport velocity, kw. This scheme required the wind speed and concentrations at 10-m elevation. The 10-m wind speed has been calculated from aircraft-measured wind speed (nominally 300-m elevation), assuming a logarithmic wind profile and the parameterization given by Wu [1995]. These wind speeds are listed in Table i  The species examined were chosen because of their relevance to photochemistry in the MBL and to check for consistency among measurements, photochemical model results, and transport models. They represent a mix of species with different aqueous solubilities, sources, and photochemical lifetimes.

03 -H202 -CH3OOH
According to current understanding, destruction of ozone in the MBL is thought to proceed through the following se-

TRACE A study region showed a similar trend with altitude but with net column 03 production near zero. This indicated the South Atlantic basin was near 03 steady state during the TRACE A period.
The key reactions contributing to net 03 production are detailed in Figure 7. CH3OO and He 2 radicals contributed equally to gross 03 production in the MBL. Above the MBL, He 2 was diagnosed as the primary source of gross 03 production. OlD-H2 ¸ and O3-0dd H reactions were the principal 03 destruction reactions with the former being greater at low altitude. In case d the predominance of O1D-H20 in the gross destruction of 03 was limited to altitudes below 2 km and in case i its predominance extended up to 7 km, as was discussed above. Gross 03 production was nearly in balance with O3-0dd H destruction at all altitudes and it is the reaction of OlD with water which shifted the balance toward net 03 destruction in these marine profiles.

NO-NO2-PAN-HNO 3
NO is required for 03 production in the troposphere and its absence or very low concentration, along with H20 , was responsible for net 03 loss in the MBL. Even though the TRACE A MBL was depleted in NO and NO2, there was still considerable movement in the oxides of nitrogen from the lower FT to the MBL and on into the ocean. This is shown schematically in Figure 8. The downward transport of peroxyacetylnitrate (PAN), which comprises most of the oxides of nitrogen flux, exceeded the rate of HNO 3 production and its transport and thermal decomposition was sufficient to maintain observed NO and NO 2 levels without having to invoke an oceanic source for NO. However, the predicted rate of HNO 3 formation from NO 2 and the entrainment of HNO 3 from aloft were significantly lower than the deposition rate of HNO 3 and inadequate to maintain observed levels of HNO 3. Figure 8 also gives lower limits for the air-sea flux of PAN and NOx. These fluxes were considered of minor importance and were estimated assuming an upper limit deposition velocity of 1 cm s-•. Duce et al., 1991]. The deposition rate of aerosol varies as a function of particle size and can be consid-   CGSFC one-dimensional model results for a TRACE A MBL simulation. The model is described by Thompson et al. [1993], but the kinetic rate constants and photophysical parameters have been updated per Atkinson et al. [1993]. erably different than a gas. Theoretically, the partitioning of HNO 3 between gas and aerosol is a function of particle mass, composition, and relative humidity. A solution to the HNO 3aerosol deposition problem was presented by Duce et al. [1991] and was similar to that used by Thompson and Zafiriou [1983]. Duce

Light Hydrocarbons, <C4
The surface emission rates of hydrocarbons were estimated from their MBL concentration, reactivity with 03 and HO, and rate of entrainment from aloft, as was done by Donahue and Prinn [1990]. In this scheme the MBL hydrocarbon species are assumed to be in steady state such that the ocean or FT supplies them at a rate equal to their photochemical destruction.

The concentration of 03 was taken from the observations and the photochemical model was used to estimate HO and NO 3 in the MBL. Hydrocarbon oxidation by NO 3 could be neglected in the TRACE A MBL. Kinetic rate constants were taken from
Atkinson et al. [1993] and evaluated assuming a temperature of 20øC and a pressure of 950 hPa. The average MBL column chemical destruction rates for C2H4, C2H6, C2H6, C3H8, and n-C4Hlo during TRACE A are listed in Table 4. The dielaverage HO concentration was --• 1.5 x 106 and the midday HO concentration was --•5 x 10 6. The oceanic sources implied by the chemical destruction rates should be reduced by an amount equal to the FT-MBL entrainment rates, which are listed in Table 4, and the differences yield refined estimates of the oceanic emission rates. For C2H 4 and C3H 6 the inferred oceanic source was an order of magnitude larger than the estimated FT source, clearly demonstrating the ocean as the source of these compounds. There was a progression in the alkanes from those indicating a predominantly oceanic source, n-C4Hlo , to those with an equal likelihood of an oceanic or FT source, C3H8, and on to those suggesting a predominantly FT source, C2H 6. Plass et al. [1992] estimated the oceanic flux of these hydrocarbons for the South Atlantic from surface seawater and lower MBL measurements (cruise track is shown in Figure 1). Their data are also listed in Table 4. The TRACE A lower-limit rate for C2H 4 was comparable to theirs, but our estimated oceanic fluxes for the other species were 2-4 times their flux estimates. Donahue and Prinn [1990] Table 4.

CH20 and Formic Acid -HFo
In Tokos [1989] proposed HFo to have a photochemical source, based upon his observations of its diel cycle in the MBL over the Atlantic in the summer of 1988. He suggested a likely mechanism was heterogeneous photochemical conversion of CH20 to HFo in cloud or aerosol, as was modeled by Chameides and Davis [1983] and jacob [1986]. Ariander et al. [1990] also noted a diel cycle in HFo and postulated that it arose from either vertical transport phenomena or sunlight-driven activity (photochemistry and biological activity). Heterogeneous photochemistry was stated by them to be unable to yield sufficient HFo. Talbot et al. [1990, 1995a] and Keene et al. [1995] have also shown that aqueous chemistry cannot explain observations of HFo in the Amazonian dry and wet seasons nor in the fall at Shenandoah National Park. Arlander et al. [1990] concluded that a more likely MBL source was the gas phase ozonolysis of alkenes, C2H 4 or C3H6, since a correlation among alkenes, HFo, HAc, and CH20 was observed. In TRACE A, however, the above diagnosed oceanic emission rates for these compounds and those from Plass et al. [1992] would be insufficient to support HFo production via their oxidation. The upper limit total oceanic emissions of alkenes (17 x 108 molecules cm -2 s -1, Table 4) were less than 1/2 of that required, 40 x 108 ( Figure 9). A viable gas phase or aqueous photochemical source of HFo remains to be demonstrated.
The available CH20 measurements within the MBL and lower FT are listed in Table 3. The CH20 instrument duty cycle, 10 min of measurements followed by 10 min of background, and the short residence time of the aircraft in the MBL, often <5 min, combined to reduce the number of MBL cases with CH20 data. It was unreasonable to perform CH20 budget calculations on so few cases.
The low CH20 concentrations were striking and significantly below ( western Australia. However, the values below the detection limit, nominally 45 ppt, for cases h, i, j, l, and m were well below these measurements. The latter measurements imply either measurement uncertainty was greater than realized or a process related to CH20 formation or loss is absent from the model.

Discussion
The simple FT-MBL and surface flux models and the photochemical point model provided a flamework from which to discuss the chemical budgets of ozone, hydroperoxides, oxides of nitrogen, and hydrocarbons. The MBL budgets of 03 and hydroperoxides appeared to be in reasonable balance. The close correspondence between measured and model values of H202, CH3OOH, and CH20 above the MBL indicated that the photochemical model chemical mechanisms and kinetic rate constants captured the odd hydrogen and odd oxygen chemistry of the FT [Jacob et al., this issue]. The 03, H202, and CH3OOH MBL measurements and model results, including MBL flux estimates, suggested that this result applied in the MBL as well. The estimated surface deposition velocity together with a mean MBL height yield pseudo-first-order loss rates of 1.4 x 10 -s s-• for HFo, HNO3, CH20 , and H202, and 6 x 10 -6 S-• for CH3OOH. The surface loss rate of H202 was approximately twice its diel-average photochemical loss rate. For CH3OOH , surface deposition was about 1/2 its dielaverage photochemical loss rate and that for CH20 was about 1/3 its diel-average photochemical loss rate. The inclusion of surface deposition in the point model reduced the instantaneous steady state values of H202, cH3OOH, and CH20 to 1/2, 2/3, and 3/4 of their values without surface deposition. Modeled and measured concentrations of H202 and CH3OOH were resolved with surface deposition, whereas model CH20 remains significantly greater than the measured values. In total, the oxides of nitrogen family was balanced in the MBL, but this required the inclusion of gas-aerosol partitioning for HNO 3 and an additional mechanism for converting PAN to HNO3 is needed. The lack of model-measurement closure in CH20 was consistent with observations at Mauna Loa Observatory and model simulations of the chemistry of this location [Heikes, 1992;Liu et al., 1992;Heikes et al., 1996]. The CH20 analytical method used during TRACE A was also used in a series of CH20 methods comparisons at Mauna Loa Observatory [Heikes et al., 1996]. It consistently gave results which were the same (within estimated instrumental precision) or higher in value than the other measurements. Hence it was considered to provide an upper limit to ambient CH20 there.
A HFo formation mechanism operative in the MBL has yet to be identified. The magnitudes of the missing CH20 sink and HFo source were comparable. While intriguing, whether the missing CH20 sink and HFo source are related and whether this relationship was a consequence of the low NO environment may have been fortuitous and cannot be firmly established here. However, the role of heterogeneous chemistry on CH20 and HFo in a cloud-impacted MBL is the subject of future work.

Conclusions
Chemical budgets were constructed for several species in the MBL and presented for ethane, ethylene, propane, propylene, n-butane, formic acid (HFo), methylhydroperoxide, oxides of nitrogen (i.e., NO, NO2, PAN, HNO3) , hydrogen peroxide, and ozone. A diel-average photochemical point model was used to evaluate local photochemical production and loss of these species. An entrainment model was used to estimate material exchange between the lower free troposphere and MBL and a resistance deposition model was used to calculate material exchange across the air-sea interface. The results suggested the ocean to be the source of measured alkenes in the MBL and that the ocean is the most likely source of the shorter-lived alkanes: propane and n-butane with a smaller contribution from the FT. The estimated hydrocarbon sources from the ocean were in agreement with prior analyses. Transport from the lower FT together with surface loss could not account for measured concentrations of HFo and HNO 3. A photochemical source of HFo is needed in the MBL. The transport of PAN from the FT to the MBL exceeds the rate of NO2 oxidation to HNO 3 and was more than sufficient to maintain observed NOx levels without having to invoke an oceanic source for NO. The total flux of the sum of NOx, PAN, and HNO 3 was in balance with the surface deposition flux of HNO 3 and indicated balance of the oxides of nitrogen family. However, the predicted rates of HNO 3 formation and HNO 3 entrainment from aloft were inadequate to maintain observed levels of HNO2 unless aerosol partitioning and depositional effects were included. The estimated dry deposition flux of HNO 3 to the South Atlantic during TRACE A was about 10 times the annual average estimate for this region and comparable to the deposition rate for all nitrate-containing species.
The destruction of 03 within the MBL was more than balanced by transport from aloft. The principal destruction process was through photochemical reactions and mediated by the formation and surface deposition of H202 and CH3OOH. A direct loss of ozone to the sea surface was of secondary importance. CH3OOH loss to the sea surface and its transport into the FT from the MBL was estimated to occur at a first-order loss rate of 6 x 10 -6 S -1 for a mean MBL height of 700 m. H202, HFo, HNO 3 and CH20 losses from the MBL are estimated at rates of 1.4 x 10 -5 s -1. Sea-salt aerosol confounds the surface deposition of HNO 3. Inclusion of surface loss improved the agreement between model-predicted and measured concentrations of HNO3, CH3OOH, H202, and CH20 , species which had been overestimated in the MBL by the photochemical point model. A strong but unknown CH20 sink was required in addition to surface deposition to resolve model and measurements in the MBL. The missing sink of CH20 and source of HFo were comparable.